Experimental Variability within Animal Assays and

Transcription

1 FDA Points-to-Consider Documents: The Need for Dietary Control for the Reduction of Experimental Variability within Animal Assays and the Use of Dietary Restriction to Achieve Dietary Control* WILLIAM T. ALLABEN, ANGELO TURTURRO, JULIAN E. A. LEAKEY, JOHN E. SENG, AND RONALD W. HART Food and Drug Administration, National Center for Toxicological Research, Jefferson, Arkansas ABSTRACT Standard protocols for conducting chronic toxicity and carcinogenicity studies have been refined over the years to carefully control for many variables. Nevertheless, over the last 2 decades, there has been a steady increase in variability, a decrease in survival, an increase in tumor incidence rates, and an increase in the average body weight of control animals among the various rodent species and strains used for toxicity testing. These observations have prompted an evaluation of chronic study designs to determine what factor(s) may be responsible for such confounding changes. Ad libitum feeding and the selection of successful breeders with rapid offspring growth is believed to be at least partially responsible for the heavier, obese rodents with which many laboratories are coping today. As a result of these changes, some studies used for the evaluation of safety have been deemed inconclusive or inadequate for regulatory purposes and either additional supportive studies have been requested and/or studies per se have been repeated. Research on the molecular mechanisms of caloric restriction and agent-induced toxicity at the Food and Drug Administration (FDA) National Center for Toxicological Research stimulated the first international conference on the biological effects of dietary restriction in 1989; this was followed in 1993 by an FDA workshop exploring the utility of dietary restriction in controlling reduced survival in chronic tests and an international conference in 1994 exploring the implications for the regulatory community of using dietary restriction in toxicity and carcinogenicity studies used in support of a sponsor s submission or in risk assessments. The outcome of that conference was the FDA s commitment to develop Points-to-Consider documents that address the issue of dietary control in chronic toxicity and carcinogenicity studies. Keywords. Caloric restriction, bioassay variability; cancer bloassay; rodent survival; rodent body weight INTRODUCTION Based on the premise that the rodent is a surrogate for humans, the biomedical research community has, over several decades, developed and used various rodent animal species and strains to identify and understand the biological mechanisms of disease, to assess the safety of food additives and drugs, and to evaluate the toxic and/ or carcinogenic effects of chemicals. During the 1960s, a large-scale hazard identification chemical testing program was developed by the National Cancer Institute (NCI) to screen numerous chemicals to determine potential cancer-causing substances (Bioassay Program). In an attempt to stabilize testing practices, standard testing protocols were established. The NCI Bioassay Program was transferred to the National Toxicology Program (NTP) in To further reduce intra- and interlaboratory variability and promote reproducibility, the NTP continued to refine and standardize testing protocols. With the development of the Food and Drug Administration s (FDA) Good Laboratory Practice (GLP) guidelines in 1978, sponsors and petitioners submit safety assessment data to the FDA that typically follow similar standardized testing * Address correspondence to: Dr. William T. Allaben, Food and Drug Administration, National Center for Toxicological Research, 3900 NCTR Road, Jefferson, Arkansas 72079; wallaben@nctr.fda.gov. 776 protocols when chronic carcinogenicity data are required by the agency. In spite of the numerous refinements to testing protocols and the establishment of regulatory testing guidelines, over the past yr, there has been a steady increase in variability, a decrease in survival, an increase in tumor incidence rates (Fig. 1 ), and an increase in the the var- average body weight of control animals among ious rodent species and strains used for toxicity and carcinogenicity testing (7, 14, 15, 21-23, 30). In fact, adequate survival of control animals (e.g., at least 50% after 2 yr on test) and/or control animal outcome variability have become such a problem for those in the pharmaceutical industry that sponsors typically load 2 separate control groups when conducting 2-yr toxicity and carcinogenicity studies (Dr. Joseph Contrera, personal communication, FDA Center for Drug Evaluation and Research Rockville MD). Poor survival in other testing programs has also been reported (8, 10, 11, 17). Moreover, the increased tumor incidence rates in control animals confounds the interpretation of agent-induced tumors at best and potentially may bias test outcome. Furthermore, variability within control populations of endpoints such as survival, background tumor incidence, and body weight often add uncertainty to safety assessments. Because a positive outcome in chronic carcinogenicity tests

2 777 FIG. 1.-Tumor prevalence in male Fischer-344 rats (as percent) for leukemia, thyroid C-cell, pituitary, and adrenal pheochromocytoma during 4 time groupings: , , , Adapted from Rao et al (21). typically triggers regulatory action or reduces the chance for regulatory approval, it was important that the agency attempt to identify the reasons for the observed drift in the preceding testing parameters and recommend corrective action. METHODS AND BACKGROUND The NTP is comprised of 3 federal agencies: the National Institute of Environmental Health Sciences (NIEHS), the National Institute of Occupational Safety and Health (NIOSH), and the FDA s National Center for Toxicology Research (NCTR). NCTR has, and continues, to conduct numerous studies under the auspices of the NTP. NCTR also assisted the NTP in developing the NTP s original toxicology data management system (TDMS) and served as the repository for that data for several years. The NCTR is the FDA s primary research arm and its fundamental and applied research activities support the FDA s 5 product centers and the Office of Regulatory Affairs. The stability of chronic toxicity and carcinogenicity studies, from within both the NTP and FDA data bases, has been of concern to the NCTR because of its own research interest, its affiliation with the NTP, and its association with the FDA s other centers and offices. The NCTR began an active research program investigating the modulatory effects of caloric restriction on disease and agent-induced toxicity in As research progressed, it became apparent that caloric, or dietary, restriction (the careful control of energy resources made available to laboratory animals) was a very powerful paradigm for delaying age-associated disease onset as well as for modulating agent-induced toxicity. The results from the NCTR s efforts and from studies in other laboratories lead to the first international conference on di- Effects of etary restriction in 1990 entitled &dquo;biological Dietary Restriction&dquo; (12), which was cosponsored by the NCTR, the International Life Sciences Institute (ILSI), and the National Institute on Aging (NIA). That conference focused on understanding the complex physiological and biochemical changes that were responsible for the life-protecting observations resulting from controlling calories in laboratory animals and what the implications were for using such a paradigm in toxicity testing protocols. It was becoming clear that there was a significant and disturbing drift occurring in chronic toxicity and carcinogenicity control animal data bases (10, 11, 23). Moreover, some sponsors submitting carcinogenic study data to the FDA had (or were planning to) incorporated some type of dietary restriction paradigm into their protocols. The NTP also designed and started select studies that incorporated dietary restriction. Because dietary restriction is such a powerful modulator of disease and toxicity, and because dietary restriction was being incorporated into more and more protocols, from which data potentially could be submitted to the FDA, the NCTR organized and sponsored a workshop on dietary restriction for FDA research and review scientists in early That workshop focused on poor survival in control animals in chronic studies, the regulatory consequences of using dietary restriction to control for poor survival, what the implications were for the review process, and how to identify studies that may have incorporated dietary restriction without a sponsor s consultation with FDA review scientists. There was rapidly emerging scientific information that identified both the importance of body weight in predicting tumorigenic potential and the great variability in body weight across test groups and among studies (26, 28, 29). Moreover, there was great interest in using dietary restriction for controlling poor survival, body weight variability, and spontaneous tumor incidence. The success of the FDA workshop, in conjunction with the preceding events, lead to the second international conference on dietary restriction in March 1994, entitled &dquo;dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies&dquo; (13). That conference, cosponsored by the FDA, Environmental Protection Agency (EPA), NIEHSINTP, NIA, American Industrial Health Council (AIHC), and ILSI, resulted in several important conclusions and recommendations (Table I). The FDA made a commitment during that conference to develop an agency document that addressed the concerns about diet (ad libitum feeding)-associated variability in chronic toxicity and carcinogenicity studies. The FDA would make the document available to other federal agencies, industry, and the general public for review and comment. The FDA formed a Dietary Control Committee, which has representation from the FDA NCTR, Center for Drug Evaluation and Research, Center for Biologics Evaluation and Research, Center for Veterinary Medicine, Center for Medical Devices and Radiological Health, Center for Food Safety and Applied Nutrition, Office of Orphan Products Development, and Office of Regulatory Affairs. This committee was charged with developing an agency &dquo;points-to-consider&dquo; document, which would make recommendations that sponsors and petitioners could follow to establish dietary control in chronic toxicity and carcinogenicity studies, that is, use of dietary or caloric restriction techniques to control body weight (growth rate),

3 - - ~~ - - ~ ~ ~ ~ ~ ~ ~ ~~ ~ ~~ ~ ~ ~ ~ ~ ~ ~~ ~ ~ - ~ ~ ~ ~- ~ ~~ ~ ~ ~ ~~ -~ TABLE I.-Dietary restriction: implications for the design and interpretation of toxicity and carcinogenicity studies-conference outcome. a International conference that was held February 28-March 2, 1994, in Washington, D.C. I thereby reducing variability and spontaneous tumor rates, and enhancing survival. NCTR scientists from the Caloric Restriction Program developed the initial draft document, and the FDA Dietary Control Committee reviewed and modified the document, which resulted in the development of 2 Points-to- Consider documents: (a) &dquo;the Need for Dietary Control for the Reduction of Experimental Variability within Animal Assays&dquo; and (b) &dquo;the Use of Dietary Restriction to Achieve Dietary Control.&dquo; The draft documents were then sent to 116 scientists from government, industry, and academia for review and comment. RESULTS AND DISCUSSION Examples of the Problem 1. Survival. Control animal survival at 24 mo has been shown to vary between 7 and 73% for male Sprague-Dawley rats (6), between 40 and 85% for male Fischer-344 rats (6), and between 40 and 97% for male B6C3F, mice (29). Besides the variability noted, these survival changes follow a timeline such that there was much higher survival in the 1960s and 1970s than we see in the 1980s and 1990s. In 1970, male Fischer-344 control rats had a mean survival rate of 80% at 24 mo FIG. 2.-Liver tumor incidence for control female B6C3F, mice in NTP chronic studies that initiated chronic dosing from 1981 to 1990 and were reported to June Each marker is a study with approximately 50 animals. (r) is a correlation coefficient, which is significant at p < on test, whereas in 1981 survival had been reduced to 60% at 24 mo on test (21). The NTP data base identifies a current low for survival in 1 Fischer-344 rat (male) study of 36% (8). 2. Tumor Incidence. As an example of spontaneous background tumor increase and variability, it has previously been shown in control B6C3F, mice, liver tumor incidence for males has varied between 10 and 76% and tumor inci- for females between 2 and 57% (26); lung dence for males has varied between 4 and 42% and between 0 and 26% for females (9). Again, the higher incidence levels are occurring in the late 1980s and in the 1990s. In control Fischer-344 male rats, pituitary tumors have increased from 9 to 26% and leukemia has increased from 5 to 55% over a 25-yr period of time (10). Expanding on previous work, liver tumor incidence and variability for control female B6C3F, mice from NTP studies initiated between 1981 and 1990 are shown in Fig. 2. The data in this figure represent all but 4 studies conducted in that time frame (103 bioassays). 3. Body Weight. Increases in the incidence of both neoplastic and nonneoplastic disease endpoints with a concomitant decrease in survival (morbidity and mortality increases) seems to be strongly associated with the profound increase in adult rodent body weights that has occurred over the past yr (24, 27). Body weight has also been shown to be directly related to survival (3, 8, 12). For instance, the trend toward decreasing survival in Sprague-Dawley (SD) rats is accompanied by significant increases in average body weight. In control male SD rats, average body weight has risen from approximately 550 g 20 yr ago to over 900 g today (8, 18). An increase in average body weight has also been reported in control male and female Fischer-344 rats (4, 11) as well as control B6C3F, mice (9). It has also been reported that there was a large variability in control group body weight for B6C3F, male mice in recent NTP studies (8). An example of this is shown in Fig. 3, where body weights ranged between 30 and 48 g. There are several possible reasons for the drift in body weight: (a) breeder selection for rapid growth and reproductive performance has led to an increased demand for energy (a type of genetic drift); (b) ad libitum feeding, which has provided the fuel for this increase in the energy demand and led to the increases in body weight; and (c) housing conditions, such as individual or group-housed,

4 779 FIG. 3.-A body weight at 12 mo (BW12) distribution for a representative NTP study in female B6C3F; mice. The study is the control group from female animals for the bioassay of pentaerythritol tetranitrate, with individual body weight for each animal represented graphically. The bar height denotes the number of animals at each 1 g weight (rounded, i.e., for a 40 g bar animals from and were included). and cage type, such as polycarbonate with bedding or wire mesh, have likely contributed to the variability in body weight (9). Dietary Control If energy demand has increased and ad libitum feeding has provided the fuel for that demand, then perhaps rodent food consumption is a good &dquo;biomarker&dquo; to control. However, studies conducted over the years with SD rats have shown a significant variability in feed consumption, averaging between 21 g/day and 32 g/day (8). The fact that most studies use group housing also makes it difficult to identify accurately the proper amount of feed to use for individual animal demands. Moreover, ad libitum feeding is so variable, as measured across individual animals, that it represents a moving target and therefore one that is difficult to control. Furthermore, there are complications with accurate measurement of food consumption, such as feed spillage, coprophagy, group housing (requiring the assumption that all animals consume the same amount of feed), activity levels, different types of feed dispensers, and feed palatability differences. Body weight, therefore, is probably the best &dquo;biomarker&dquo; to control to reduce high background tumor incidences, body weight variability, and increased rodent survival at the end of chronic studies (24 mo on test) (29). Work initiated by Turturro, Duffy, and Hart (26) and continued by Seilkop (24), Haseman (8), and Turturro (27) has demonstrated that body weight selected at a period of time before the expression of tumors, such as body weight at 12 mo on test, correlate very well with the development of tumors at the end of 24 mo-in control animals. Idealized Body Weight Growth Curve Maintaining an agreed-upon body weight through the use of dietary restriction, and therefore achieving a pro- FIG. 4.-An idealized body weight curve that predicts liver tumors in male B6C3F, mice to be about 15% at the end of 2 yr on test. NCTR AL (top long-dashed line) is a plot of the mean body weight for ad libitum-fed animals; NCTR CR40 (bottom long-dashed line) is a plot of the mean body weight for NCTR s 40% calorically restricted (60% of ad libitum-fed) animals. The solid line in the middle represents the idealized body weight growth curve for male B6C3F, mice that predicts a spontaneous liver tumor incidence of ca. 15% at age 26 mo (24 mo on test). The small dotted lines represent standard deviation ± 5%. A study with the hypnotic drug chloral hydrate is currently ongoing at NCTR, which establishes an idealized body weight curve through the use of dietary control (5, 20). jected, acceptable, background tumor incidence for control animals, will reduce variability around a population body weight mean and reduce the variability of results within and across studies. As a consequence, and dependent on the body weight selected, animals will be healthier, live longer, and exhibit a lower incidence of various disease processes, including cancer. It is critical that body weight curves be selected that permit the expression of common tumors (thus demonstrating that appropriate mechanisms responsible for the expression of carcinogenesis are functioning) but do not compromise the study through poor survival. This factor should be adjusted against the sensitivity of the organism to the compound as reflected in the establishment of a maximum tolerated dose (MTD) and through the results of subchronic and supportive mechanistic studies. Because it is important for each animal to receive the measured amount of food daily, individual animal housing will be required. Figure 4 demonstrates an idealized body weight growth curve for male B6C3F, mice that predicts a background liver tumor incidence of about 15%. Thus, the biological mechanisms responsible for the reduction and/ or expression of liver tumors are operative (bioassay sensitivity maintained) while body weight and tumor incidence variability are predicted to be significantly reduced and survival enhanced. A study is currently being conducted at the NCTR, under the auspices of the NTP, which is achieving an idealized body weight curve that has a predicted background tumor incidence in control animals of about 15~0 (5, 20).

5 780 TABLE II.-Selected comments from Points-to-Consider reviewers. Supportive Studies Given the effects of body weight on toxicity, supportive and mechanistic studies, especially those used to determine the MTD, should incorporate similar levels of dietary control. Type of animal housing (caging) used can also affect body weight levels, and supporting studies for a chronic test should utilize similar housing conditions as the animals on the chronic test. The efficacy of dietary control (achieved through the implementation of dietary restriction) is not an all-ornone phenomenon but, rather, represents a gradient of responses directly related to the extent of caloric intake (19). Most studies that have incorporated dietary restriction have shown that even relatively small levels of dietary restriction can alter an organism s response to chemical exposure (l, 2, 4). These results have led to 2 concerns: (a) ad libitum feeding often results in rodents being overfed, which leads to poor survival and high background tumor incidence with increased morbidity, making interpretation of test outcome difficult in not impossible (18) and (b) dietary restriction will result in an underfed, less sensitive animal test model (16). It should be stressed that care must be taken to optimize caloric intake in a manner that enhances survival and reduces spontaneous disease but does not compromise the ability to detect toxic or carcinogenic endpoints. The use of dietary restriction to achieve dietary control in animal assays, while scientifically defensible, requires scientific judgment and careful analysis due to the action of dietary restriction on a number of key biochemical and physiological mechanisms, each of which modulates the expression of toxicity (1, 18). Fortunately, there have been a number of studies that have utilized dietary restriction and provide guidance in the design and utilization of this paradigm in toxicity tests (5, 16, 20, 25, 31). It is believed that the use of dietary control will: (a) reduce drift in the body weight of study animals, (b) standardize the mean weight curves of rodent populations on test, (c) stabilize the impact of body weight on the determination of a MTD, (d) adjust for the effects of body weight to reduce high background tumor incidence and thus enhance survival, and (e) reduce variability in body weight, background tumor incidence rates, and survival. Points-to-Consider Reviews Eighty-three of the 116 scientists agreed to review the document, 21 were unable to meet the time requirement or were unavailable, and 12 did not respond. The committee has received and reviewed 68 reviews. Table II lists selected comments from those that were asked to review the FDA Points-to-Consider documents. While some of the comments in the table may sound negative, well over 60% of the reviewers supported the need for dietary control in rodent bioassays. Reviewers addressed policy issues as well as scientific, technology, and cost of accomplishing dietary control in chronic bioassays. Each document is currently undergoing revision. The agency plans to prepare final draft documents and make them available for general comment prior to their formal publication. CONCLUSIONS The lack of dietary control has left uncontrolled perhaps the most significant modulator of toxicity that has yet been characterized. The resultant variability of test outcome confounds the assessment of efficacy, toxicity, risk, and risk assessment/risk-benefit analyses. It is important to employ standards for animal tests that provide consistency and equity in interpretation. To evaluate nonneoplastic as well as neoplastic disease outcomes, it is important that this factor be controlled. The FDA Pointsto-Consider documents recommend that sponsors and petitioners consider variability in outcome that results from ad libitum feeding and other factors influencing body weight, such as multiple housing, feeder type, and cage type, and take appropriate steps to control these variables to improve the interpretation and regulatory utility of studies conducted in support of submissions, through dietary control and the establishment of idealized body weight growth curves. ACKNOWLEDGMENTS Some research projects reported in this paper were supported in part by interagency agreements between the NCTR and the National Institute of Aging for the Project on Caloric Restriction and between the FDA/NCTR and

6 781 the National Institute of Environmental Health Sciences/ National Toxicology Program. REFERENCES 1. Allaben W, Chou M, and Pegram R (1991). Dietary restriction and toxicological endpoints: An historical review. In: Biological Effects of Caloric Restriction, L Fishbein (ed). Springer-Verlag, New York, pp Allaben W, Chou M, Pegram R, Leakey J, Feuers R, Duffy P, Turturro A, and Hart R (1990). Modulation of toxicity and carcinogenicity by caloric restriction. Korean J. Toxicol. 6: Berg T and Sims H (1960). Nutrition and longevity in the rat. 3. Longevity and onset of disease with different levels of food intake. J. Nutr. 71: Chou M, Pegram R, Gao P, and Allaben W (1991). Effects of caloric restriction on aflatoxin B, metabolism and DNA modification in Fisher 344 rats. In: Biological Effects of Dietary Restriction, L Fishbein (ed). Springer-Verlag, New York, pp Leakey, J (PI) (1996). Chronic bioassay of chloral hydrate in male B6C3F, mice using idealized body weight curves that are normalized by modulation of caloric intake; NCTR Protocol ; studies started 1/ Hart R, Keenan K, Turturro A, Abdo K, Leakey J, and Lyn-Cook B (1995). Caloric restriction and toxicity. Fund. Appl. Toxicol. 25: Haseman J (1992). Value of historical of historical controls in the interpretation of rodent tumor data. Drug Info. 26: Haseman J (1995). Statistical considerations in long-term dietary restriction studies. In: Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies, R Hart, D Neumann, and R Robertson (eds). ILSI Press, Washington, D.C., pp Haseman J, Bourbina J, and Eustis S (1994). Effect of individual housing and other experimental design factors on tumor incidence in B6C3F, mice. Fund. Appl. Toxicol. 23: Haseman J, Huff J, Rao G, and Eustis S (1989). Sources of variability in rodent carcinogenicity studies. Fund. Appl. Toxicol. 12: Haseman J and Rao G (1992). Effects of corn-oil, time-related changes and inter-laboratory variability on tumor occurrence in control Fischer 344 (F-344/N) rats. Toxicol. Pathol. 20: International Conference (1990). The Biological Effects of Caloric Restriction. March 5-7, 1990, in Washington, D.C. 13. International Conference (1994). Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies. February 28-March 2, 1994, in Washington, D.C. 14. International Life Sciences Institute (1993). The Biological Effects of Caloric Restriction, L Fishbein (ed). Springer-Verlag, New York. 15. International Life Sciences Institute (1995). Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies, R Hart, D Neumann, and R Robertson (eds). ILSI Press, Washington, D.C. 16. Kari F and Abdo K (1995). The sensitivity of the NTP bioassay for carcinogen hazard evaluation can be modulated by caloric restriction. In: Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcirzogenicity Studies, R Hart, D Neumann, and R Robertson (eds). ILSI Press, Washington, D.C., pp Keenan K and Soper K (1994). The effects of overfeeding and dietary restriction on Sprague-Dawley rat survival and early pathology biomarkers of aging. Toxicol. Pathol. 22: Keenan K and Soper K (1995). The effects of overfeeding and moderate dietary restriction on Sprague-Dawley rat survival, spontaneous carcinogenesis, chronic disease and the toxicologic response to pharmaceuticals. In: Dietan, Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies, R Hart, D Neumann, and R Robertson (eds). ILSI Press, Washington, D.C., pp Kritchevsky D and Klurfield D (1987). Caloric restriction affects Am. J. Clin. Nutr. 45: experimental mammary gland tumorigenesis Leakey JE, Turturro A, and Hart R (1996). Idealized weight curves to control background tumor incidence in B6C3F, mice. The Toxicologist 30: Rao G, Haseman J, Grumbein S, Crawford S, and Eustis S (1990). Growth, body weight, survival and tumor trends in F344/N rats during an eleven-year period. Toxicol. Pathol. 18: Rao G, Haseman J, Grumbein S, Crawford D, and Eustis S (1990). Growth, body weight, survival, and tumor trends in (C57BL/6 X C3H/HeN) F1 (B6C3F1) mice during a nine-year period. Toxicol. Pathol. 18: Roe F (1994). Historical histopathological control data for laboratory rodents Valuable treasure or worthless trash? Lab. Anim. 28: Seilkop S (1995). The effect of body weight on tumor incidence and carcinogenicity testing in B6C3F, mice and F-344 rats. Fund. Appl. Toxicol. 24: Sheldon W, Bucci T, Hart R, and Turturro A (1995). Age-related neoplasia in a lifetime study of ad libitum-fed and food restricted B6C3F, mice. Toxicol. Pathol. 23: Turturro A, Duffy P, and Hart R (1993). Modulation of toxicity by diet and dietary macronutrient restriction. Mutat. Res. 295: Turturro A, Duffy P, and Hart R (1995). The effect of caloric modulation on toxicity studies. In: Dietary Restriction: Implications for the Design and Interpretation of Toxicity and Carcinogenicity Studies, R Hart, D Neumann, and R Robertson (eds). ILSI Press, Washington, D.C., pp Turturro A and Hart R (1991). Longevity Assurance mechanisms and caloric restriction. Ann. N.Y. Acad. Sci. 621: Turturro A and Hart R (1992). Dietary alteration in the rate of cancer and aging. Exp. Gerontol. 27: Turturro A and Hart R (1994). Modulation of toxicity by diet: Implications for response at low-level exposures. In: Biological Effects of Low Level Exposures: Dose-Response Relationships, E Calabrese (ed). Lewis Publishers, Chelsea, Michigan, pp Witt W, Sheldon W, and Thruman J (1991). Pathological endpoints in dietary restricted rodents Fischer 344 rats and B6C3F, mice. In: Biological Effects of Dietary Restriction, L Fishbein (ed). Springer-Verlag, New York, pp